Nebius Group N.V.

The Nevis Group's first quarter 2025 earnings call. My name is Neil Doshi. I'm head of investor relations. Joining me today to discuss our results are Arkady Volosh, founder and CEO, and members of the management team. Our remarks today will include forward-looking statements, which are based on assumptions as of today. Actual results may differ materially as a result of various factors, including those set forth in today's earnings press releases. and in our annual report on Form 20F filed with the SEC. We undertake no obligation to update any forward-looking statement. During the call, we will present both GAAP and certain non-GAAP financial measures. A reconciliation of GAAP to non-GAAP measures is included in today's earnings press release. The earnings press release, an accompanying shareholder letter, and an investor presentation our website at group.nebius.com forward slash investor-hub. We ask that you enter your questions into the webcast portal, and we will be reviewing and consolidating the questions for Q&A. And now I'd like to turn the call over to Arkady and the team, who will go over a few slides that we've presented for this investor presentation. Yeah, thanks.

Thanks, Nils. Thank you, everyone, for joining our Q1 2025 results call. I will start with saying that demand for AI compute was very strong in first quarter and actually our results show it. Our revenue grew nearly 400% year over year and our analyzed run rate revenue grew nearly 700%. We saw great momentum in our core infrastructure business. We ended the quarter with a solid cash balance of $1.4 billion. and we actually continue to invest in our infrastructure. To that point, we are rapidly building our capacity to serve customers around the world. This is a global race, as you understand, and we are well placed with our footprint in the US, Europe, and now in the Middle East. As you can see here on the slide, we added three new locations recently, and there's more to come. We're exploring new locations for capacity build-out and we hope to share more news with you very soon. We also announced some new partnerships this quarter to build our strong relationship with NVIDIA as well as Metal Llama. And finally, we had a very productive quarter with respect to building out our technology stack. And we are getting industry recognition for our AI cloud offering. Here, as an example, SEMI Analysis ranked us in the golden tier of the GPU cloud cluster marks rating system. And now I'll hand over to Ray Kuralenko to discuss some of the key products we launched in Q1.

Thank you, Arkady. And hello, everyone. We believe that the And the quality of our tech is a significant differentiator against the other Neo clouds. We made the great progress in Q1 in further developing our AI cloud offering and had a number of notable product launches. First of all, we launched the Slurm-based cluster upgrades, such as automatic recovery for the failed nodes, proactive system health checks to detect the issues before jobs actually fail. This changes reduces downtime for customers and improved capacity availability on our infrastructure, which led to around 5% improvement on available nodes for the commercial use, which is quite significant. Several platform services were released and moved from the beta phase to the general availability. MLflow and JupyterLab notebook as examples, but there was much more. We also invested a lot of time and efforts in reliability and performance of the platform. Notably, we launched enhanced object storage, and this ensures that large data sets can be assessed and saved quickly during model training runs, reducing time to results. And building on that foundation of our homegrown storage capabilities, we have also partnered with three leading storage providers, such as DDN, VAST, and VECA. And that enables us to deliver the best possible experience for all customer scenarios going forward with the Blackwell generation clusters. Last but not least, we expanded integrations with external AI platforms, such as Metaflow, D-Stack, Skypilot, and that allows customers to bring the existing tools into our ecosystem with minimal friction. And partners, Daniel will talk about our partners.

Thanks, Andre. In addition to strengthening our product in Q1, we also made significant progress towards expanding our partner ecosystem. From further building out our data storage solution portfolio, as Andre mentioned, with industry leaders, we extended our core AI cloud capabilities to the ISV landscape with tight technical integration. And we made announcements enabling customers to consume Nebius infrastructure across a wide segment of the industry. Equally important are the relationships we have with the full range of AI marketplaces, established channel partners that help us meet the customer demand for our AI infrastructure across the globe. I'd also like to talk about NVIDIA. As you know, NVIDIA is an investor in our company, and we have a long history of working with the NVIDIA team, and we want to continue to build on that relationship. In Q1, we made several announcements with them. Point one, In the Q1 timeframe, Nebius and NVIDIA announced that Nebius would be one of the first AI clouds to offer the NVIDIA Blackwell Ultra AI factory platform. We also became a launch partner for NVIDIA Dynamo, one of the most efficient solutions for scaling compute during inference. And Nebius was also named one of five reference platform NVIDIA cloud partners in this timeframe, helping us as we specialize and deliver AI accelerated services built on NCP reference architectures. And finally, some breaking news. Nebius will support the NVIDIA DGX Cloud Leptin Marketplace at launch. We couldn't be more excited with our partnerships, not just with NVIDIA, but across the landscape. So with that, I would like to hand it over to Roman.

Yeah, thank you, Daniel. Let's speak a little bit about customers. Our strategy is to serve a wide variety of customers with our robust platforms. We have hundreds of customers, both managed and self-service, who use Nebios Cloud Platform for training and inference workloads across various industries, such as tech, media and entertainment, life science, and more. With our expanding capacity footprint and global sales support, we are now able to serve customers 24-7 This truly tailored approach of our high-level experts on both sides of the Atlantic that together with the advanced software platform goes beyond commoditized GPU as a service offerings. This highlights our flexibility and ability to rapidly adapt to the evolving needs of our diverse customer base while delivering high quality solutions powered by our tech stack. This is what our customers value the most. They recognize that we are building an AI specialized cloud with hyperscaler level of capabilities. Actually, all those factors contributed to our strong Q1 results. And going forward, the demand environment for AI compute remains robust. And Our sales momentum has continued into Q2. April's analyzed run rate revenue was $310 million, and we are continuing to experience strong demand into May. And now I'll pass it to Tom Blackwell to walk through our guidance.

Yeah, thanks very much, Roman. And so, you know, as Roman said, we've had, we've had a great start to the year. Very, very strong first quarter. And we're carrying in strong momentum into the second quarter. So we feel very confident in our ability to achieve the ARR guidance for the whole year that we gave, which was 750 million to a billion. We're well on track to achieve this. So we're also reiterating our overall revenue guidance for the group, which is in the range of 500 to 700 million dollars. So we're turning to profitability here. So we're maintaining our adjusted EBITDA guidance for the full year. So just to elaborate on that a bit, while we expect adjusted EBITDA to be negative for the full year, we plan to turn positive at some point in the second half of 2025. On CapEx, we're currently planning CapEx for approximately $2 billion for 2025. This is a bit up from the previous guidance of one and a half billion due to a couple of factors. First, we had some capex spend that had been planned for late Q4, which actually fell in early Q1. So some of that leads to the increase towards $2 billion. And also, we've always said that we want to be opportunistic when it comes to really ramping up our infrastructure capacity as we see demand. And so we want to be able to secure that demand well. And so we've considered some additional investments beyond the initial data center expansion plans. For example, you may have seen some coverage recently around the data center in Israel, which we think is a great opportunity. It's a great market. And we'll come on and give some more color on that later in the call. So looking to the midterm, this is a great business. It's in a great industry. And we think the future opportunity is immense. When we look at the midterm, we believe that this business will achieve mid single digit billions of dollars in revenue. And we're actively building out our capacity pipeline to support that scale of revenue growth. The reality is that there are also scenarios where we could grow more aggressively. And so Andrea and his team are very focused on really building out the whole infrastructure potential pipeline that would enable us to deliver potentially more than a gigawatt of capacity in the midterm. So if we do that, that would obviously, that would allow us to achieve significantly more revenue than the kind of mid-term guidance that we're talking about here. So we'll be optimistic and we'll go after opportunities as we see them. Some of the factors that could drive that additional incremental growth on top of the mid-term guidance is as we see more adoption from enterprise level customers and also potential sort of larger, longer term contracts. And again, Arkady will give a bit more color on that at a later stage. In terms of profitability, you know, this is a business that we can grow profitably. And we anticipate medium-term EBIT margins to range in the sort of 20 to 30% range. So this will be supported by our AI cloud business reaching scale. We also have, you know, we have an important differentiator, which is the full stack and particularly the software at the top end of the stack. And the software is, it's a very important part of our business model. What makes us attractive to clients, sticky to clients. And ultimately we think it's what's going to allow us to achieve higher margin, create higher margin business models, and really service customers in different ways and in a wider range of customers that allow us to basically get increased that effective revenue per GPU. So not just the GPU as a service model, but it's a broader range of sort of revenue sources. So we also, I think it's also important to note that we actually take a very conservative view on depreciation. So actually with all of these numbers, we apply here a four year depreciation schedule. while others I think typically use more of a five or a six year depreciation schedule within our industry. Longer term, I think while we see 20 and 30% as the EBITDA margins in the midterm, longer term, we could go beyond that. I think there are a number of scenarios as we continue to scale up and expand the business where we could go well north of 30% in the longer term. So just to wrap up, we're building AI infrastructure successfully in its scale. I think as you've heard Akari talk about on previous calls, fundamentally, we think our differentiation and what sets us apart really comes down to two things. Above all is the quality of our technology. There's also our access to capsule that's allow us to take advantage of that technology and to ramp up and to scale up quickly. So briefly on the technology, we have an amazing team of engineers who are building amazing hardware, software, and services. These engineers, they're really, they're the best of the best in the industry. It would take years to build a team of that quality, and we're really proud to have them. They're building great tech. We're building out our native AI cloud, and we're expanding the range of AI native customers that we're able to service. And really, it's the AI cloud that we build, it goes well beyond what you might call a classic bare metal offering. We're building out strong partnerships, as Daniel talked about, within the ecosystem, and all of this is allowing us to reach and service a broader range of customers. In terms of capital, so we think that we're actually in a very favorable position and actually quite a unique position among Neoclouds to really finance this future growth in an efficient way. So we have significant capital funding potential for the core business, which actually comes from our various ownership and equity stakes of non-core businesses. And the monetization of these potential equity stakes can really translate efficiently into bottom line results of the core business. So just to give some examples of what we're referring to here, you may have seen ClickHouse in the news lately. We have a 28% or roughly 28% minority stake in the business, and this can potentially be a very important source of future capital. So according to some of the recent press reports, there's a fundraising round underway at the moment, which would potentially value the business at around $6 billion. And we believe that business will continue to perform extremely well and grow significantly from current levels. We have Taloka. And we're extremely pleased to announce the arrival of the emergence of strategic investors from Jeff Bezos and Michal Perachin coming into the structure. And we think that their investment involvement in the business is really going to help Taloka to scale up among the top tier of AI data companies globally with great backing from these investors. And by what's important for us, we think this is great for Taloka, but it's great for us as well because we, and for our shareholders, because we maintain a significant majority economic interest in Taloka. So we'll benefit from all of the upside. We also have AB Ride, one of the best autonomous vehicle teams in the world. They're doing great this year. In the last quarter, we've announced they've entered into partnerships with players like Uber, Hyundai, Grubhub, Rakuten. These partnerships really underscore, I think, the strength of their tech and the team and places them really among a select group of global leaders in their field. A brief note on AB Wright, as we've mentioned previously, we're in fairly active talks with potential third-party investors and strategic investors that could come into the business that we believe would really help them to scale up even faster and really build them to build their businesses. But again, while we would have always looked to retain significant economic interest in the upside. So it's really our ability to use these assets and these stakes, which gives us a really very attractive source of financing, So when we think about the future, you know, sort of billions and dollars of investment in the core business, we'll be able to very effectively monetize these businesses and to grow extremely efficiently in a way that really minimizes any dilution to existing shareholders while allowing us to stay very disciplined in terms of debt. So just again, just to sort of summarize, you know, once we achieve adjusted EBITDA profitability, our strong balance sheet and continued low interest burden, we believe will allow revenue growth to translate very efficiently into bottom line results.

I'll stop there, and Neil, I'll hand back over to you for Q&A. Great. We'll pause for just a moment just to collect some of the questions. Great. All right, so let's start with our first question. You just guided to midterm revenue and margins. What do you mean by midterm, and what are the building blocks to get it there?

Roman? Yeah, thank you, Neil. Our base case plan calls for several billion dollars of revenue in the midterm over the next few years. While our base case assumes that we grow our capacity to support this type of revenue goal from 2025 levels of 100 megawatt, our ambition is to grow much larger and much faster. For that, we are building a data center pipeline to provide scalability to more than one gigawatt of power. Also, as Tom said earlier, how quickly we get there will be a function of how fast we can scale and capture demand through more enterprise level customers and longer term contracts. And also a few words about the margins. Our target of 20, 30-bit margins is a function of a few factors. First, great mix of workloads where we can run our GPU fleet with a high level of utilization for a longer period of time. Second, I would say, is software. we put a lot of efforts into developing our software, which allow us to assume contribution from high margin software and services revenue over the long term. And worth to mention, in addition, we take a more conservative view on depreciation, where we use a four-year depreciation schedule, while others use five or six year. So when more work flows shift to inference, this will come to higher margins for us as well. Great. Thanks, Roman.

And Roman, maybe you could take the next question too, which is around Q1 ARR was ahead of what you discussed on the last earnings call. What really drove that strength? And how are you feeling about the full year?

Yeah, as I said, overall demand environment in Q1 was strong. Customers wanted access to GPUs. And we see that demand strengthened each month. Customers, I believe, recognize the value of our infrastructure and software. We were able to provide reliable and scalable service. Our software enables customers to start accessing clusters with thousands of GPUs just within a matter of days and not weeks. And we heard from some of our core customers recognition of that. We also saw the benefits of our sales team ramping up and especially the investments in our pre-sales and solution architects and customer success team. Now we can provide 24 seven wide globe support. And I believe it like significantly contributed to improve our sales process and obviously post-sale customer success. Our brand awareness is also growing. We put a lot of effort there and also thanks to industry recognition, for example, seminar analysis, cluster marks, gold status that Arkady mentioned contributed. And we see that our pipeline becomes more deep and strong. Also, we see that our approach to bring the newest chips online as early as possible like not responding to specific contracts, but in this like more cloud manner and our flexibility to provide the real cloud terms, the combination of pay as you go or reservations of different length is paying back as well. A good example was the DeepSeq moment in February. We could very quickly respond to the peak demand to NVIDIA H200 chips that we had deployed in more volumes than maybe some other players at that moment. All that resulted in a strong growth and we reached a record high number of our managed customers during Q1. A few words about the full year. We continue to see a solid start of Q2. The demand remains robust. April annualized run rate revenue of 310 million confirms that, and we are seeing the strong momentum continue to move. In the second half of the year, we expect to bring Blackwells for customers, which should provide further support to our revenue profile and gives us confidence that we can deliver on our guidance of $750 to $1 billion analyzed revenue by the end of Q4 2025. Great. Thanks, Roman.

So you discussed getting to positive adjusted EBITDA margins by the end of the year. Can you provide an update when you think that'll happen? Maybe we'll go to Tom for this.

Yeah, sure. So I guess I touched on this briefly in the presentation, but just to pick up. So, you know, first of all, achieving positive adjusted EBITDA is an important milestone for us. And it really highlights that we're very focused on getting to profitability. And as we set out in some midterm targets, we believe this is a business that can post really strong profitability going forward. So specifically, again, with respect to adjusted EBITDA, we intend to reach positive territory at some point during the second half of the year. One thing I would note is actually if we break it down and look at the core infrastructure business, then we'll move in faster there and we'll get to positive adjusted EBITDA probably sometime in the third quarter. The next goal will obviously be to then focus on reaching positive adjusted EBIT, and we're working full steam towards that goal.

Great. And Tom, maybe sticking with you, you know, there's a question here about CapEx. We've raised the CapEx guidance. Can we provide any update on the reasons for this?

Yeah, sure. I mean, you know, so our primary business model is predicated on building capacity for demand. And we've been very fortunate to be able to finance a lot of our CapEx with our cash on hand up until now. So, you know, looking at this year, so first of all, in terms of the kind of the specific guidance for this year, As I mentioned earlier, we had some CapEx spend that had been keyed up for the end of the fourth quarter last year, which got pushed into the first quarter. That's just down to sort of typical quarter to quarter fluctuations based on sort of various factors related to data center build-outs. But, you know, we, again, you know, We want to be opportunistic. You know, we view the targets that we've set out as base cases. But, you know, there are a lot of scenarios where we can do more and go more aggressively. And when we see an opportunity to do so, that's in a way that's value accretive to our shareholders. We want to be able to do so. So, for example, you know, again, when we see an opportunity to ramp up capacity faster around the existing demand that we can see. We want to be able to do so. So again, the Israel data center is one that we haven't initially had in our roadmap, but it was an opportunity that came along and we felt that was a good one for us to go for. And so we're very pleased. And I think probably at a later stage, I'll ask perhaps Roman to give a bit more color around that. But it's a great market and we're very excited to be getting into that market. So it's a new geography on top of some of the previous geographies that we've been focused on. But in terms of those incremental data center buildouts like Israel, from a revenue standpoint, we'll be investing, putting the capacity in place later this year. And the revenue will be, we'll see more contributing to 2026, you know, keeping us very much on that path towards the sort of mid-single digit billions of revenue that we spoke about earlier in the presentation. Great.

Maybe, you know, keeping with the theme of CapEx, Tom, I know you touched a little bit in the slides on how we're going to finance our future growth. So how do we expect to finance the CapEx expansion, given that the cash balance now is below what we're planning to spend?

Yeah, sure. So just to kind of recap, so in fact, if we look at Q1, so we've already spent $544 million in the first quarter towards that overall $2 billion CapEx that we're talking about. And At the end of the quarter, we have 1.44 billion of cash remaining on the balance sheet. So we forget about our ability to finance that capex. And also, again, I just come back to this point. So, you know, kind of going beyond that and looking further afield, you know, the equity stakes that we have in these non-core businesses, again, we believe will provide us very significant funding sources for future investments. Again, so we can continue to ramp up and scale beyond this year in ways that really minimize, again, dilution to shelters and allow us to stay very disciplined on that. And I think it's a really important point. Obviously, as a public company, we have access to more traditional funding sources, and we will look at those from time to time when we believe they make sense and are value-inclusive. And I guess just another point that I would make is that, again, given... that we have right now, we have very low debt and we anticipate continuing to have relatively low levels of debt. So that means that we're gonna be able to reinvest a significant amount of our revenue back into driving value creation in our core AI infrastructure business.

Great. Looks like we're getting a kind of a higher level question here, which is around future growth. So maybe Arkady, maybe you can tell us where you're seeing the growth in the future in this business.

Well, our current customer base, the majority of them, they are all those new AI companies that emerged in the recent couple of years and actually continue coming to the market every month. They are very advanced in AI technology. We call them AI natives. They are smart. They are very fast growing. Actually, we are like them and they like us. Those companies are usually venture backed and understandably the majority of them are in the US. That's why we're so focused on building our data center capacity in the US right now. All the growth you currently see in Nebios, all these quarter results, year results, most of our revenues, they come mostly from this market now. The second very promising sector which, by the way, is not in our revenue yet, is all those frontier AI labs, big customers. We haven't tapped this market yet, but we're doing a lot to be ready to serve them and help them to grow faster. In order to serve these, those customers, we will need much more and much bigger data centers. And we are actually getting ready for this. This is why we have our pipeline to get to this more than 100 gigawatts of data center capacity. So we are not there yet, but we will be there soon. That's the second sector. The third sector, maybe the most promising sector in terms of growth, is the enterprises. AI technology today has reached just a small fraction of the corporate clients. Everybody talks about it. But at the same time, this is where the world expects the majority of added value which AI will be creating. And by the way, our full-stack solution and high-level services which we provide is very much relevant exactly here. This market is much more global by nature because the real industry, real enterprises are everywhere in the world, many countries. And this is where our European and global infrastructure presence will be in high demand, I think. And by the way, outside of just several hyperscalers, we are one of very few AI cloud providers that actually can serve corporate clients in multiple geographies. So this is the most promising sector for us, I believe, in near future. And there's also the fourth sector, which we're also watching carefully. It's a potential market among national AI projects. We hear more and more about them. And here again, we see a huge opportunity for us. And we plan to build our AI factories in different countries and geographies, in the US, in Europe, in the Middle East, and actually elsewhere. But all in all, those four sectors and the whole market is just the beginning for AI technology, for AI business. AI infrastructure will be in high demand in many industries and in many geographies. And Navios will be there to serve this demand. That's great. Thanks, Arkady.

All right. Let's see. Next question. How does Toloka deconsolidation impact your business? I can probably take this. So Toloka is our AI data solutions provider. They've done a really good job in terms of building their business. They have high quality customers like Amazon, Anthropic, Microsoft, Poolside, Wecraft, and Shopify. And we believe Toloka really has great growth prospects. And this is really validated by their investments from the Bezos Expedition and Mikhail Karakian, the CTO of Shopify. You know, given these growth prospects, we're happy to retain a significant majority economic stake in Toloka. And, you know, as we now our voting shares drop below 50 percent, we will be deconsolidating Toloka. Since the transaction closed in May, we will be updating our financials and guidance ex-Toloka in our Q2 earnings report. All right, looks like we have a few questions around infrastructure. Maybe we can start with Andre on this one. Andre, can you provide us an update on your capacity expansion plans for this year? Sure, Neil.

We are aggressively building and acquiring, generally, our data center capacity and expanding the footprint. What we announced in February was the New Jersey Data Center, and this is a built-in suite project built by partner according to our specifications and design and this is quite important because of it helps us to deliver the efficiency of the power usage and the cooling efficiency again what we what we demand from ourselves we expect that the first capacity in Jersey to be operational in the late summer and I'm that will continue to roll out on a periodical basis in conjunction with the demand. We also announced the Kansas City. And the first part of it is already fully operational. And that was the last deployment of the Hopper GPU generation for us. And at the moment, there are black holes deployed in the second part of the Kansas City, and they will be available in the platform a bit later in the second quarter. We also announced and actually launched Iceland, which is fully operational at the moment, and our build-out in Finland is going quite well and exactly on track, and we expect the first phase of the expansion will be operational in late in Q3, and the second phase will follow it closer to the year end. And I expect that operationally we'll have over 100 megawatts of capacity deployed this year. Great.

Thank you, Andrey. Let's see, can you share more about the new site in Israel? And can you discuss your expansion strategy beyond the EU and US? Tom, I know you kind of alluded to this. Maybe, Roman, you can talk a little bit more about this and elaborate on Israel.

Yeah, thank you, Neil. So first of all, launching in Israel means for us that we opened one more market. Like Andrei said, we were very much and continue to be focused to scale our capacity in Europe and US, but we don't want to be limited only by those markets. So first of all, it's a new market in India. Israel has a great AI market with a lot of AI native startups, enterprises, and R&D centers of the global corporates. This is a great aid for us to a lot of customers. But what's also important that this is our first, but probably not the last step in supporting national AI factories. And we hope to support and build more national AI factories around the world. And we'll look to see how we can plug into those initiatives across Europe, Middle East, and the rest of the world. open, opportunistic, and looking to this market, as Arkady mentioned. Great. Thanks, Ram.

All right. Can you share an update on your GPU rollout plan for this year? Andre?

So aside from the Israel capacity, we are very much on track with the rollout that we planned earlier. This year we deployed, in Q1 specifically, we deployed the Hopper's generation H200 specifically. At the moment, we are rolling out the Blackpools, as I already mentioned, and they will be available on the platform shortly. And we also started to deploy the Grace Blackpool family, so the GB200 family. We expect that in Q3, the Blackwell Ultra generation will start with the first deployments. And the majority of this year, we will be actually deploying the Blackwells.

Great. Thank you, Andrey. Andrew, maybe sticking with you again, we have a question around regulatory issues around tariffs. Any thoughts on kind of the impact of tariffs on our data center expansion plans? And also just how are you thinking about the cost to our business?

Good question, Neil. There's definitely was in this a certain, I mean, It's not clear around global tariffs, but based on where we stand now, we don't believe that the current status would result in major changes to our expansion plans. We also believe that we can navigate through the current environment of tariffs without significant impact to our costs. I would, however, It's just a very dynamic situation, and things can change quite quickly, as we already saw during the Q1, and we are actually monitoring the situation. Great. Thanks, Andrey.

So it looks like we have some questions about customers. So maybe I'll give this to Daniel. Daniel, tell us more about Nebius' customers and why are they choosing Nebius over other providers?

Great. Thanks, Neil. And thanks for the question. First of all, our customers choose us because we offer a high performance, resilient and scalable alternative for other cloud providers. But what really makes a difference, our differentiation lies in our deep expertise and hyperscale infrastructure in our role as a hands-on practitioner along with our customers. So we're not just another platform vendor. What this does is ultimately enable us to drive a greater return for every AI dollar our customers spend. Some examples of that in Q1, we saw great momentum and new wins in vertical industries like healthcare and life sciences, like media and entertainment and financial services. One customer of ours, Captions, is a leading AI video platform. They partnered with us to scale GPU training for the next generation audio to video model Mirage. And so by leveraging our infrastructure, they accelerated their time to market. They empowered their creators to deliver emotionally compelling and story-driven content and ultimately push the boundaries of AI-powered storytelling. So a great example in the media and entertainment industry. Another example would be Quantory. They're a top biopharma company partner of ours, they used Nebius to build a framework for 3D molecular generation. And ultimately by increasing the amount of molecules that they could model, they achieved chemically valid structures and really enabled faster scalable R&D and accelerated the innovation that they have in drug and materials discovery. So really monetizing and unlocking the power for AI for those customers. And that's just the beginning. So looking ahead, And we're doubling down on the verticalization of AI solutions across the enterprise. Customers ranging from retail to robotics, as they embed AI deeper into their core operations, we want to be right there with them to drive measurable results.

Great. Thanks, Daniel. We're getting a few questions around contracts. So maybe, Roman, can you tell us a little bit more or give us an update on what type of contracts we're seeing in the market, maybe in terms of structure and duration?

Yeah, thank you, Neil. The first thing I want to highlight that the benefits of coming to Nebios is our flexibility that allow us to support and grow with the native AI tech startups and meet their needs in flexibility. Contract lengths tend to go from several months and to year and beyond. In addition, as we are just starting to bring the fleet of Blackwells, that opening up more discussions about longer-term contracts. New generation, high interest, and GB200s and GB300s. We expand this to drive more demand and give us flexibility on the types of the contracts we will be able to secure. Great. Thanks, Roman.

All right. Question around NVIDIA. Can you talk a little bit more about the NVIDIA relationship? How's that progressing? Daniel, I know you shared some thoughts in your slide, but anything more you want to elaborate there?

Yeah, I think between Andre and I, we've covered a lot. I'll do a little bit of reiterating here just in case anybody missed a few details. Obviously, we have not just a tight collaboration, but a longstanding collaboration with NVIDIA. They have been an investor in a capital raise last December with us. And we have a very robust go-to market that we've built with them. In Q1 in particular, like I mentioned before, across the Blackwell family, but particularly as we announced the Blackwell Ultra AI factory platform, we are going to be one of the first vendors to stand up the GB300 and VL72 powered instances. And we think this is going to be a real game changer in the market. And we're right there with NVIDIA as those roll out. We also talked about the ecosystem and NVIDIA Dynamo, this open source inference framework. And so as we continue to roll out the scale and the variety of the AI factories that are needed in the market, we're right there with them in real time. And then the other thing that I mentioned earlier, but it's still important, is The ability to stand up a cloud based off of NVIDIA architecture that actually performs to spec and delivers at least a dollar for every dollar invested, if not more, is what the NVIDIA Cloud Partner Program is all about and what the reference architectures that we're one of five partners really delivers for customers. It's a clear validation of our technical leadership. And that just rolls over into the marketplace that they're standing up with DGX Cloud, Leptin, and lots of other opportunities, whether it be with the startup community or expanding out and helping enterprises monetize AI. We've been very much in lockstep with NVIDIA for a very long time and look to have a very bright future. Great.

Thanks, Daniel. We're definitely getting some questions around our software stack. We also get these questions quite a bit when speaking with shareholders and investors. So, you know, it seems like we've launched a lot of products in Q1 on the software side. And how does our software stack compare with our competitors? And what really were kind of the biggest launches? Andre, maybe you want to take this.

Yeah, Neil, we started Mambios with a clear goal to build a full stack AI cloud. That means from the day one, our focus has been to create a software stack that is specifically built for the AI workloads. And our stack is basically three layers. The first one is the layer that manages our hardware. And since we designed our hardware, we also can offer the tools to monitor its performance and optimize the usage. The second layer is that we build a full cloud platform. It's pretty similar to the big hyperscalers. It's a virtualized environment, so customers get more flexibility and better stability overall. And the third layer is an application layer where you can deliver the pre-configured third-party AI tools and that simplify the entire AI development process. So we ship quite a lot of products in Q1. I think it's around 50 products across the AI Cloud and the AI Studio. Notably, as I was saying on the opening, we launched this learn-based cluster upgrade such as automatic recovery, proactive system checks, issue detection before the actual jobs fail. And these changes significantly reduce the downtime for the customers and improves the time to recovery on our side. We made a lot of efforts and made a lot of improvements on our object storage. Now, just we boost the speed of read and write per compute node. That, again, ensures that the data sets can be assessed and saved quickly, quickly enough for the training runs, and that improves time to results during training. Also, the partnership with the leading storages companies helps us to provide more flexibility to our customers. And as I said earlier as well, the integrations, we believe that it's very important to integrate with existing AI platforms such as Metaflow, DStack, and SkyPilot, and that just allows the customers to bring their jobs and their tools, existing tools, on us with minimal friction.

Great. Thank you, Andrei. What is it, so in terms of financial performance tying software back to our finance, how does that drive revenue and margins? Tom, maybe you can take a stab at this question.

Yeah, sure. I mean, so I think it's important to understand that our software stack, it's a critical part of the offering, right? So many of our customers rely on the stack to help them manage and execute their workloads. And again, we think we're relatively unique in having that full stack offering within, you know, among the Neo clouds in our space. It also just makes us very sticky with customers. So some of the things that, you know, for example, it allows us to do, it allows us to provision large clusters of GPUs quickly so customers can start their jobs faster. We've created various tools to help them manage their data, models, and track their progress. And so, I mean, when we think about revenue contribution, I mean, I suppose the revenue contribution that you could break out as of today, it's relatively small. But, you know, it's really very much an added value part of the offering. And again, it just drives overall customers coming to us and drives overall revenue. And so it's part of the broader offering. And, you know, we're going to be very focused on building it out, building out use cases and continuing to make the products more sticky with customers sort of going forward. So I think over time, it can be, it probably can become even a more significant standalone driver of high margin revenue. But it's really about helping us access a wider range of customers, offer high margin services, high margin products, keeping them on the platform. And so it's an important part of our overall revenue growth kind of going into the midterms.

Great. And, you know, coming back to the funding question, you know, just maybe a little bit more pointed here. So, you know, the question is, you will need funding for this year, but also for the coming years. And how are you thinking about financing options? You know, I know, Tom, you talked a little bit about that on the slides and kind of in the CapEx question, but Anything else you want to kind of reiterate on this point?

I'll just actually kind of reiterate quite briefly on this. Again, it's important for us as we think about funding the growth, we want to do this in a way that minimizes shelter dilution and allows us to be prudent in terms of debt. We're in a great position to do that with the cash we have on the balance sheet and with the potential to monetize these various stakes in the non-core businesses. We'll, of course, in due course, we'll be considering other more classical opportunities through the capital markets, and we'll update us when we have more. But again, we feel very good about our ability to continue to fund this growth based on the available sources of capital that we have to us.

Great. There's a question on the other business, specifically AVRIDE. You said you may explore strategic options. Can you maybe share a little bit more about why you're excited about AVRIDE and maybe what those options could potentially be? Arkady?

Well, yeah, we covered that several times. There's just a few independent autonomous vehicle platforms that can compete on the US market today. Definitely, the market is excited about what Wayman has achieved. Everybody sees it. And one of the very few players that can actually build a platform comparable to that. As you can see, other big market players actually recognize this. Look at the recent announcement of a variety of partnerships with Uber, with Hyundai, with other big players. And as we said, they need to grow. They need to grow much faster. It's yet another capital intense business in our portfolio. And we are in active discussions. We can confirm that. We're in active discussions with potential strategic partners and investors who can actually really help to drive growth to this ambitious project. Great.

Maybe one last question, a clarifying question. Can you explain exactly what you mean by midterm? Tom, maybe you want to take this one.

Yeah, sure. So again, just to recap on that. So as we look into the midterm, we really believe this business can scale up quickly and achieve sort of single, mid-single digit billions of dollars of revenue. So what we mean by midterm, I mean, with effectively a few years. But at the same time, as we've tried to sort of outline, we're working very hard to go as aggressively as we can, and we'll get there as soon as possible. And we have, I think Arkady sort of framed how we're thinking about the future growth. And so again, a lot of our existing revenues and forecasts stems around this kind of AI native customer base. But I think Akari set out the other sort of incremental sources of growth around the enterprise customers, the big labs, and so on and so forth. So in our mind, we think midterm is a few years, and we'll go as quickly as we can.

Great. All right. Thank you, everyone, for participating on our first quarter 2025 earnings call. And we will see you again on our Q2 call. Thanks. Thank you.